 Okay, everybody, we're back. This is Dave Vellante. I'm with Wikibon.org and I'm with my co-host Jeff Kelly. John Furrier's going to run a crowd chat, so check out the crowd chat.net and you'll see what John's up to. We're here live at the Vertica user event in Boston, Massachusetts. We're at the Western Hotel in the waterfront. It's been a great two days. A lot of practitioners, really a focus, has been on use cases, practitioners, applying technology to create business advantage. Earlier today, we heard George Kadifa layout, HP's vision for its software business and Haven. The big piece of that was really the large market opportunity and Jeff Kelly and I have talked extensively about our premise at Wikibon, which is that big data practitioners are going to be creating more value than the actual technology community and we're seeing what George Kadifa called 100X opportunity to create value and monetize data and that's really what it's all about. But underneath it all is a technology. Technology enables the people and processes to be built up to create that value and Shilpa Luwande is here. She's the Vice President of Software Engineering at Vertica. She also is responsible for customer support within HP Vertica and she is a long time Vertica employee. She was there from the beginning. Shilpa, welcome to theCUBE. It's great to see you. Thank you so much. Thanks for having me. Yeah, you're welcome. So take us back, you know? I mean, I remember the early days of Vertica. When I first met them it was probably like 2006, 2007. So you guys had been at it for a while but so you were early on the team. Yes, I started Vertica when we were, before the first line of code was written. We were three people. Myself and another person started on the same day and we didn't even have whiteboards in the rooms and we had three rooms, really, a three office. It's like a pre-start-up. Like a little garage type of a thing. Yeah, we were in a little, we were in this brick facility where everybody else was doctors' offices and kids running around and stuff like that. Before we knew our data center was so hard that the building couldn't take what we were subjecting it to. So we were kicked out. So take us back to the early days. What were you guys seeing? What excited you? Why did you join this tiny little idea? What got you there? What drew you in? Yeah, so I found them quite by accident, actually. And then I went and met Mike Stonebreaker. And Mike said to me that, well, I used to work at Oracle for a long time, eight years or so. And Mike said, you know what? I have no idea if this is going to work out, how it's going to work out, but at the least you would have built something really cool, a really cool piece of technology. You would have built a solution from ground up, a database for ground up. So as for somebody who had been a database engineer, worked at a very large database corporation, this was a really cool opportunity, just as an engineer. And it's amazing being here at the first Vertica user conference to see how far we've come and what all the cool stuff people are able to do with the technology we built. From a technology standpoint, I mean there's a couple of paths you can take as a starter. You can say, okay, I want to take an existing technology, I'm going to make it a little bit better and maybe a little bit faster and I'm going to go compete. You can say, I'm going to try to do something completely different and think out of the box. You see this a lot out of Israel, the innovation that comes out of Israel, you go, wow, how'd they think of that? So Vertica's more the latter, right? So talk about what's different fundamentally from what everybody else was doing at that time. Right, so fundamentally all the other existing database technology that was out there had been built for a different problem. So it had been built to solve OLTP problems and so when the problem domain changed to analytics and data warehousing and so on, what people started to do was they just retrofitted other things around their existing platforms to make that work and so you have materialized views and cubes and all these things that came about tried to improve performance for large scale analytics but really fundamentally they were dealing with an architecture designed for something else and so they didn't really have the freedom to really take tackle the hard problems there. So with Vertica what we chose to do was that we looked at what is the workload we're trying to address, what are the fundamental architectural changes that are needed and we also took into account the hardware innovations that was happening, right? So disks are getting bigger and all that but IO bandwidth is not growing at the same rate but CPUs on the other hand are getting faster and faster and more and more cores. So we wanted to shift the burden from the disk to the CPU and so a number of these things sort of lined up for us to create Vertica. That's what we fundamentally changed. Views compression, views column stores as well as an MPP scale out architecture. So all of these elements made us truly unique. So the profound sort of technical innovation was as you say shifting the burden to the CPU which the price was dropping, performance was going through the roof. Commodity hardware essentially. Okay. Being able to run on commodity hardware really take advantage of the enormous amount of compute capacity that you can get by scaling out. So talk about how that differentiates you from some of your competitors out there who still reliance on kind of proprietary somewhat expensive hardware and the ability to scale out versus scale up. What does that give you? Is that mainly a cost advantage? What kind of contrasts that with what you see some of your competitors doing? So it gives us a number of advantages. So first of all Vertica has this real flexibility of being able to be deployed on different configurations. So whether it's an appliance Vertica actually has a so smaller footprint of the software. It's only a 50, 70 megabyte software image. So you can actually run Vertica embedded inside appliances and things like that for applications. But you can also run it in the cloud and so on. So we have this enormous flexibility of deployment but the scale out architecture really allows us to scale limitlessly. So we have customers who are in the petabytes. We have many, many customers who are in petabytes. I would say that we probably run some of the largest analytics installations out there right now. And that fundamentally changes also your cost equation. When you are trying to grow and your data volumes are unpredictable, you're trying to give more and more users in your organization access to analytics. You want the ability to cheaply, incrementally scale that solution. And so having that fundamental architecture of a scale out architecture and commodity hardware-based architecture gives us those advantages. George talks about 100% of the data. You heard his keynote this morning. We're going to capture 100% accommodate, allow ISVs to accommodate 100% of their data or developers. What about the architecture and were you guys thinking about this at the time? Allows you to accommodate 100% of the data, the diversity of data. Was that by design? Was it just lucky? Talk about that a little bit. So certainly, Vertica by itself, by the fact that we built it from ground up, we've architecturally designed it to be very modular. So over time, we have been able to expand it from just being a very fast database to really a more comprehensive analytic platform. So we've over, I would say, starting from the release five onwards, we have started to put in APIs everywhere through the database platform. So we have now extensibility on types of analytics you can do. We have extensibility into sources that you can plug into Vertica to load the data and so on. And so that is where I believe that gives us the most, you know, the power to really be able to handle all the data over time. So it's a matter of building plugins, connectors, leveraging some of the other HP components like autonomy idle, ArcSight and things like that, as well as our extensibility, the ecosystem of partners. 700 connectors. Connectors and the ecosystem of partners you can build on that. So, you know, making the APIs be open as well as be performant, right? So it's fine to provide extensibility if, you know, the things that you write as plugins don't perform at the same level as software we would build. And it's not that useful, but we've taken great care to make sure that the APIs we put out really give you, you know, performance as close to possible for engine. When you set out the change, the sort of direction of the industry, which is really what you were doing, you always make trade, you have to make trade-offs. And, you know, you start going out and selling to early customers and they say, can I do this, can I do this? And they ask you questions in the context of the existing marketplace. And you have to be very disciplined and say, well, no, we purposefully don't do that. This is what we do. And we're, you're trying to be proactive and get ahead of where the market trend is. And sometimes you can be too early. Sometimes you can be too late. You guys happen to hit it very nicely. But what were some of those trade-offs that you guys made that you discussed that you purposefully said, okay, no, we're not going to do that. We're not going to get lured into that because that will sacrifice our future. Yeah, so early on we, you know, so anytime a customer would come to us and say, oh, we would like one database that does everything. Can we have one database standardized across the organization that does everything? So we said, no. If you want to do transaction processing, Vertica is not the thing for you. So we do not want it to be used in that fashion. And then there were other use cases like possibly serving platforms, right? Where people wanted to use a website and a database that's backing a website that millions of users are going to access. We didn't want to go against that as well. So there are a number of complementary technologies that over time we've worked with or our customers have used. Our real focus has been on analytics, right? Complex analytical workloads that really make it unique. Now, you said early on in 2005, you were observing from the hardware trends the obviously the price performance improvement of microprocessor based technology continues today. But there's another hardware trend that's emerging that I want to ask you about which is Flash. At the time Flash was, you knew it was coming but it was so expensive there was no way you were going to leverage it back then. But now 2013 and beyond, it looks like in many cases Flash is going to be less expensive than certain types of disk. How does that play into your thinking? Do you not need Flash? Is it something that you can leverage going forward? What are your thoughts on that as an architect? Yeah, so what we look at is we've built into Vertica of the flexibility of being able to use multiple types of storage. So you can tear your storage that you run Vertica on. So let's say you're the newest data, the newest 30 days of data, for example, could run on Flash and the rest could run in disk. You could run Vertica completely in memory if you wanted to by just giving it a lot of RAM. So our architecture is, for us, all of these storage types, we are fairly agnostic to them. Now, in terms of performance, you would, with a given amount of Flash, you could do a lot more with Vertica than you would need, than the amount of Flash you would need to say speed up a row store like Oracle. So we are looking at use cases right now. There are some use cases where we find Flash is useful but perhaps it's not as necessary as you might need it in other environments. So. So Shilpa, so we've heard a lot about Haven this morning here at the show. And obviously Vertica plays a huge role in the platform. Can you talk a little bit about the process of integrating all these different technologies that HP had, the big data assets really that they had in their hardware assets as well and integrating those into a kind of a seamless platform. From an architect's perspective, what were some of those challenges and just kind of walk us through how that process came about and how that's progressing? Yeah, so HP certainly has a number of unique assets and they are very complementary to what Vertica does. So ArcSight, for example, has a lot of machine data, connectors and so on. Autonomy has a lot of access to truly unstructured data like audio, video images and so on. So we are working on creating seamless connectivity between these so that through Vertica you could reach out to some of these other data sources and grab that data and then make it available for analysis through SQL. So we're looking at it in two ways. One is providing the Vertica user base, which is primarily a SQL oriented user base, giving them access to these other data sources, perhaps to enrich the other information that is already stored in Vertica. But we're also looking at it from, like let's say the autonomy user base for them to be able to augment their analysis with some of the data that is stored in Vertica. So whatever interface is natural to them. So it's a matter of really understanding what the different usage patterns are, what is natural to the user base and trying to make underneath the covers, plumbing the systems together to make it seamless. So walk us through a little bit of that kind of the plumbing underneath. What were some of the challenges really of doing that and bring those different types of technologies together to make them work seamlessly and allow really the end users to get the best of both Vertica autonomy and some of the other technologies? Yeah, I think the real challenge is that the systems are separate products. And so what is the real challenge is ensuring that there is a use case that truly benefits from having multiple products. So what we do not want to do is have one big monolithic sort of an offering where you have to take all of these pieces together. So we want to keep them as fairly separate products and have a loose integration but a more integrated user experience. So that's where we are going towards. So it's kind of a modular approach. A modular approach. So you can pick and choose the components that you need for your particular problem that you're trying to solve. So if you're trying to, for example, combine the structured data about your customers with audio records from their call center logs, you can use Vertica autonomy. If you're trying to combine Vertica with, as a let's say an extended analysis store to augment ArcSight security, same capabilities, then you just use those two components together. So having spoken to a lot of customers here this week, one of the things that we've heard a couple of times from, several times from customers is that they're very impressed with some early customers as well of Vertica with the kind of the technical chops, if you will, the engineers and the architects and really the technical team when they would come out on site and kind of help them set things up and kind of get things rolling. I think that's largely in part to your work, but as you expand and you're growing and you're part of HP, how does that influence kind of how you grow the team and kind of keep that focus on really expert technicians and engineers and how do you kind of keep that culture and that eye on the prize, if you will, of that technical expertise? Yeah, so we've always had a very high bar for hiring at Vertica and we continue to keep that. Vertica has really a great cashier right now in the technical community, so we are able to attract a lot of talent, so we continue to grow in that fashion. And culturally, I think our focus has always been to make customers successful, so we try and make it very easy for our developers to hear directly from customers and so on, like in this conference, you'll see a whole bunch of developers are here, for example, and that's what I think makes it possible. We try to still keep the close-knit feeling across the many organizations within Vertica, the engineering organization, the field organization, the support team, have them all communicating and talking still as we grow. That's what's going to keep us continuing to innovate, like we've always done. Very good, so one other thing, of course, we've heard a lot about is the need for, HP's got to reach out to the developer community to really build these applications, kind of the end at the end of Haven. There's quite a large ecosystem around Vertica right now, with a lot of partners here, we've had Tableau on and some others. Talk a little bit about, you mentioned kind of what the API is and making it open and accessible to partners. Maybe you could expand on that a little bit. What are some of the key areas that you need to do as an organization to kind of reach out to those partners from a technical perspective? Yeah, so I think we have a big focus on building a community around Vertica. So we've recently completely revamped our online community, trying to put out more and more technical content, but really trying to get our user base to contribute content there. We also have a GitHub repository for Vertica Extensions, where there's a whole bunch that we've contributed, looking for people to contribute. So really I think our focus is to expand the number of people who are contributing content and best practices and use cases and so on amongst each other, really have a larger conversation going. We also started down the effort of having formal certifications and things like that, that really when our customers are looking for Vertica professionals, there is a way for them to understand what the skill sets are and so on. So it's a part of growing the Vertica community, both informally and formally. So just generally as a technologist you look out, I mean the database world, before you started at Vertica, around that time even, let's say early 2000s, pretty boring market, right? You had Oracle had sort of done its thing, you had DB2 going along, you had SQL server, and then you had sort of the spate of innovators, like Vertica, come on, and then they got snapped up quickly because large companies like HP saw them and said, oh, we need this, that's good, this is disruptive. And now you're seeing all this other disruption with no SQL databases and open source databases. So as an observer of this industry, what comments and observations would you make about the innovations that are going on in the database and data management side of the business? So certainly I think we consider ourselves part of this whole ecosystem of open source data management and so on. We consider Hadoop, for example, as a really cool technology that is very complementary to Vertica. There is definitely some effort going, it's interesting that Mike Stonebreaker at one point wrote this very controversial article about how people who do MapReduce should learn a little about people from the database community, and there was a lot of hoopla about that for a while, but now you see that. Yeah, what does he know? Yeah, now you see that people are sort of recognizing sort of what he was intending to say there, it wasn't about one being bad and one being good, but people are trying to incorporate some of the learnings from 30 years of database research, and SQL is a good language, it's by itself not the bad thing, right? And so, and there's a number of existing skill sets out there that it leverages. So I think the no SQL community is sort of recognizing that it isn't about throwing the entire thing away, it is not the interface that's the problem, it's making the rest of the user experience work with that. So I see it's good that there's innovations going on in the open source community as well, about the no SQL community trying to get them to also have SQL interfaces to some of these products and so on, so that's a good thing. So by inference, that would suggest that the traditional sort of database architectures will evolve and as will the newer ones, they'll sort of come together and maybe reach some kind of equilibrium, but it seems like in listening to George, talking to him on theCUBE, that you're betting essentially HP strategy from the standpoint seems to be that all the growth, all the growth, but most of the growth is in this new area, this unstructured data, this diverse data, this machine data, any type of data, not so much the transaction processing. So number one, number two is the value created from that those initiatives will be far, far greater, 100x greater. So that would suggest there's some new players that might emerge in this space. HP, I'm sure you would want to be one of them. So what do you think? Do you think equilibrium, where the oligopoly just subsumes everything, guys like Oracle buy companies and act like they invented the new stuff, or do you actually feel that there will be some new players emerge, whether it's HP or somebody else that are dominant in this business? It's hard for me to predict the future, but I always feel like there's got to be a continuing spate of new players and innovation if only to keep us all on our toes. So I think that Varika is also doing some very cool stuff that about trying to make it easier for people, for example, to add unstructured and all kinds of data easily into the database. So sort of removing the barriers that people feel like, the reason why people had thought or felt like they had to invent no SQL because what databases couldn't do, right? So we are working towards making that process easier too. So it's going to be a nice competition, it's always good to have that, because that's what keeps us, keeps it fun and challenging. So let's talk about your role a little bit. How was your role kind of shifting or evolving as kind of Varika continues on this journey and is now part of HP as one of the original employees really had to help build this database? How was your role kind of evolving at the organization? So I continue to run the product engineering team, but I'm also responsible for the customer experience, so which is a great thing for me because I get to have the whole end-to-end responsibility. So I run the support services training and so on. And so it's a great way for us to completely close the loop in the sense of getting feedback from customers and incorporating that into the product and so on. So I really enjoyed that. And in terms of, I'm curious to understand a little bit more about how you mentor kind of some of the younger folks you might bring on. I mean, what are some of the things that you focus on when you bring on some new engineers, developers? What are some of the things, the knowledge you pass on to them or some of the best practices you try to instill in them? Yeah. So we have a very collaborative culture at Varika. We make it, so we have a lot of incredibly smart people, but they don't wear their brains on their sleeves or whatever you call them. You will see people are very humble, eager to share knowledge. So when a new person joins, we pair them up with a mentor, we make sure that there is somebody to hold their hand. But at the same time, we do expect our new hires to contribute right away. So we will give them stuff to do. They will, most of our new hires will end up writing features in the next release and so on. So it's a pretty inclusive as well as challenging environment. Excellent. All right, Shilpa Luande. Thank you very much for coming on theCUBE. It's always a pleasure seeing you and getting your deep perspectives on architecture and industry directions. Congratulations on all your success and we wish you the best in the future. Thank you. Thank you for having me. Keep it right there, everybody, and we'll be back. Jeff Kelly and I, we've got all kinds of stuff going on this afternoon. We're going to be running for quite some time. Actually, the conference will wind down, but we're going to have the likes of Kurt Monash on. If you don't know Kurt Monash, he's a real innovator in the quote-unquote analyst, blogging business, really knows his stuff. Vinny Mershandani, former Gartner analyst and current author will be here. So keep it right there. This is Dave Vellante with Jeff Kelly and John Furrier. We're live from the Vertica Conference in Boston. We'll be right back after this word.